www.futurenviro.es | Abril-Mayo April-May 2021 69 Smart Water | Smart Water por los técnicos correspondientes, que son los encargados de proponer las medidas correctivas o preventivas adecuadas. El algoritmo trabaja con redes convolucionales (con algoritmos de Deep Learning) y mezcla visión por computador y robótica móvil, aprendiendo patrones de comportamiento. Cada vez que se carga un video en la herramienta, los puntos se detectan categorizados en defectos estructurales o de reducción de capacidad hidráulica y se identifican a partir de unos ejemplos observados previamente. Uno de los productos de la inspección son los informes, que indican: 1. Código de defecto 2. Descripción del defecto: presencia de sedimentos, juntas desplazadas, roturas, derrumbes, cambios en las secciones, otros. 3. Ubicación en el tramo del colector 4. Calificación de la condición estructural de la red ordenada por tramos del colector evaluado La siguiente imagen ha sido extraída del informe de inspección de un colector gestionado por Canal de Isabel II: El algoritmo recoge información que permite actualizar el inventario de la red y va aprendiendo o generando una base de datos para evaluar otras inspecciones en el futuro, ayudando en la priorización de las actuaciones preventivas de la red de alcantarillado y ahorrando costes significativos. Drones autónomos para la inspección de alcantarillados visitables Para evitar los riesgos inherentes a las inspecciones de los colectores visitables y sistematizar la detección de problemas en los mismos, Canal está desarrollando drones de inspección donde la principal característica radica en que vuelen de forma autónoma, lo que reduciría al mínimo la presencia de personal en el interior de los colectores durante las inspecciones. Esta característica viene impuesta por la dificultad de volar un dron en un espacio muy reducido, bien desde la superficie, bien desde el interior. El sistema se basa en una flota de drones UAV (sistema de aeronaves no tripuladas) para la toma de datos durante toda la inspección, tras la que se analizarán los resultados obtenidos haciendo uso del algoritmo de IA descrito previamente. El objetivo del proyecto es variado, permitiendo conseguir las siguientes mejoras: Each time a video is loaded into the tool, faults are detected, categorised into structural or hydraulic capacity reduction defects and identified based on previously observed examples. One of the products of inspection are reports which indicate: 1. Fault code 2. Description of the fault: presence of sediment, joint displacements, cracks, collapse, changes in sections, other. 3. Location in the sewer section 4. Rating of the structural condition of the network in order of the sewer sections assessed. The following image is taken from an inspection report on a sewer managed by Canal de Isabel II: The algorithm collects information to enable updating of the network inventory and learns or generates a database to evaluate future inspections, thereby facilitating the prioritisation of preventive actions in the sewer network and enabling a significant reduction in costs. Autonomous drones for inspection of accessible sewers In order to avoid the inherent risks associated with inspections of accessible sewers and to systematise the detection of problems, Canal de Isabel II is developing inspection drones that can fly autonomously. This will minimise the presence of personnel inside the sewers during inspection operations, which is a key a requirement due to the difficulty of flying a drone in a very small space, either from the surface above or from inside the sewer. The system is based on a fleet of unmanned aerial vehicles (UAVs) or drones, which collect data throughout the inspection process. The results obtained are then analysed using the AI algorithm described above. This initiative has a number of different objectives and it will afford the following benefits: - Guarantee the safety of operators - Enable continuous, accurate data collection - Systematise analysis of incidents detected The final product will be a fleet of drones capable of planning and carrying out inspection tasks in accessible sewers, which are confined indoor spaces. These drones will be capable of flying in autopilot mode and making decisions during the flight without the input of operators in the sewer or on the surface. They will be equipped with physical protection against impacts and with obstacle detection and avoidance systems. The drones will also be equipped with additional elements to carry out data collection (lighting system, camera or video recorder, 3D mapping as it proceeds along the sewer), to ensure equipment recovery in the event of a drone failure and to ensure flight stability during the inspection operation. Of particular relevance are additional elements such as geopositioning systems, alarm systems, LED display signals and return home systems (which return the UAV to the inspection starting point in the event of failure). A market study found that no commercially available equipment had all the required features, though many of the components sought were available separately or partially. For this reason, Canal de Isabel II is leading a Pre-commercial Procurement (PCP) procedure. This innovative process is made up of three remunerated phases in which companies compete to develop the drone with the required features. Product development confidentiality is ensured throughout the entire process. The product delivered at the end of each phase is evaluated on the basis of 28 pre-established technical and
RkJQdWJsaXNoZXIy Njg1MjYx